wrw/sp/util.py

   1 from . import cons
   2
   3 def findnsnames(el):
   4     names = {}
   5     nid = [1]
   6     def proc(el):
   7         if isinstance(el, cons.element):
   8             if el.ns not in names:
   9                 names[el.ns] = "n" + str(nid[0])
  10                 nid[:] = [nid[0] + 1]
  11             for ch in el.children:
  12                 proc(ch)
  13     proc(el)
  14     if None in names:
  15         names[None] = None
  16     else:
  17         names[el.ns] = None
  18     return names
  19
  20 class formatter(object):
  21     def __init__(self, out, root, nsnames=None, charset="utf-8", doctype=None):
  22         self.root = root
  23         if nsnames is None:
  24             nsnames = findnsnames(root)
  25         self.nsnames = nsnames
  26         self.out = out
  27         self.charset = charset
  28         self.doctype = doctype
  29
  30     def write(self, text):
  31         self.out.write(text.encode(self.charset))
  32
  33     def quotewrite(self, buf):
  34         for ch in buf:
  35             if ch == '&':
  36                 self.write("&amp;")
  37             elif ch == '<':
  38                 self.write("&lt;")
  39             elif ch == '>':
  40                 self.write("&gt;")
  41             else:
  42                 self.write(ch)
  43
  44     def text(self, el):
  45         self.quotewrite(el)
  46
  47     def attrval(self, buf):
  48         qc, qt = ("'", "&apos;") if '"' in buf else ('"', "&quot;")
  49         self.write(qc)
  50         for ch in buf:
  51             if ch == '&':
  52                 self.write("&amp;")
  53             elif ch == '<':
  54                 self.write("&lt;")
  55             elif ch == '>':
  56                 self.write("&gt;")
  57             elif ch == qc:
  58                 self.write(qt)
  59             else:
  60                 self.write(ch)
  61         self.write(qc)
  62
  63     def attr(self, k, v):
  64         self.write(k)
  65         self.write('=')
  66         self.attrval(v)
  67
  68     def shorttag(self, el, **extra):
  69         self.write('<' + self.elname(el))
  70         for k, v in el.attrs.items():
  71             self.write(' ')
  72             self.attr(k, v)
  73         for k, v in extra.items():
  74             self.write(' ')
  75             self.attr(k, v)
  76         self.write(" />")
  77
  78     def elname(self, el):
  79         ns = self.nsnames[el.ns]
  80         if ns is None:
  81             return el.name
  82         else:
  83             return ns + ':' + el.name
  84
  85     def starttag(self, el, **extra):
  86         self.write('<' + self.elname(el))
  87         for k, v in el.attrs.items():
  88             self.write(' ')
  89             self.attr(k, v)
  90         for k, v in extra.items():
  91             self.write(' ')
  92             self.attr(k, v)
  93         self.write('>')
  94
  95     def endtag(self, el):
  96         self.write('</' + self.elname(el) + '>')
  97
  98     def longtag(self, el):
  99         self.starttag(el, **extra)
 100         for ch in el.children:
 101             self.node(ch)
 102         self.endtag(el)
 103
 104     def element(self, el, **extra):
 105         if len(el.children) == 0:
 106             self.shorttag(el, **extra)
 107         else:
 108             self.longtag(el, **extra)
 109
 110     def node(self, el):
 111         if isinstance(el, cons.element):
 112             self.element(el)
 113         elif isinstance(el, cons.text):
 114             self.text(el)
 115         else:
 116             raise Exception("Unknown object in element tree: " + el)
 117
 118     def start(self):
 119         self.write('<?xml version="1.0" encoding="' + self.charset + '" ?>\n')
 120         if self.doctype:
 121             self.write('<!DOCTYPE %s PUBLIC "%s" "%s">\n' % (self.root.name,
 122                                                              self.doctype[0],
 123                                                              self.doctype[1]))
 124         extra = {}
 125         for uri, nm in self.nsnames.items():
 126             if uri is None:
 127                 continue
 128             if nm is None:
 129                 extra["xmlns"] = uri
 130             else:
 131                 extra["xmlns:" + nm] = uri
 132         self.element(self.root, **extra)
 133
 134     @classmethod
 135     def output(cls, out, el, *args, **kw):
 136         cls(out=out, root=el, *args, **kw).start()
 137
 138     def update(self, **ch):
 139         ret = type(self).__new__(type(self))
 140         ret.__dict__.update(self.__dict__)
 141         ret.__dict__.update(ch)
 142         return ret
 143
 144 class iwriter(object):
 145     def __init__(self, out):
 146         self.out = out
 147         self.atbol = True
 148         self.col = 0
 149
 150     def write(self, buf):
 151         for i in range(len(buf)):
 152             c = buf[i:i + 1]
 153             if c == b'\n':
 154                 self.col = 0
 155             else:
 156                 self.col += 1
 157             self.out.write(c)
 158         self.atbol = False
 159
 160     def indent(self, indent):
 161         if self.atbol:
 162             return
 163         if self.col != 0:
 164             self.write(b'\n')
 165         self.write(indent)
 166         self.atbol = True
 167
 168 class indenter(formatter):
 169     def __init__(self, indent="  ", *args, **kw):
 170         super(indenter, self).__init__(*args, **kw)
 171         self.out = iwriter(self.out)
 172         self.indent = indent
 173         self.curind = ""
 174
 175     def simple(self, el):
 176         for ch in el.children:
 177             if not isinstance(ch, cons.text):
 178                 return False
 179         return True
 180
 181     def longtag(self, el, **extra):
 182         self.starttag(el, **extra)
 183         sub = self
 184         reind = False
 185         if not self.simple(el):
 186             sub = self.update(curind=self.curind + self.indent)
 187             sub.reindent()
 188             reind = True
 189         for ch in el.children:
 190             sub.node(ch)
 191         if reind:
 192             self.reindent()
 193         self.endtag(el)
 194
 195     def element(self, el, **extra):
 196         super(indenter, self).element(el, **extra)
 197         if self.out.col > 80 and self.simple(el):
 198             self.reindent()
 199
 200     def reindent(self):
 201         self.out.indent(self.curind.encode(self.charset))
 202
 203     def start(self):
 204         super(indenter, self).start()
 205         self.write('\n')